Ternary and many-body components

Intercollisional dip. In most examples of the rotovibrational spectra shown, one notices a fairly well developed dip at the Q transition frequency. 

Furthermore, sometimes a much less pronounced absorption dip is seen at the rotovibrational transition frequencies. Knowledge of the dip is nearly as old as collision-induced absorption itself the earliest report [129] mentions an unexplained component X at about 4100 cm-1, observed in hydrogen-rare gas mixtures. Subsequent studies [120, 121, 175] pointed out the main features of the new phenomenon. Specifically, it was noted that 

Later studies showed the same phenomena in deuterium and deuterium-rare gas mixtures [335, 338, 305], and also in nitrogen and nitrogen-helium mixtures [336] in nitrogen-argon mixtures the feature is, however, not well developed. The intercollisional dip (as the feature is now commonly called) in the rototranslational spectra was identified many years later see Fig. 3.5 and related discussions. The phenomenon was explained by van Kranendonk [404] as a many-body process, in terms of the correlations of induced dipoles in consecutive collisions. In other words, at low densities, the dipole autocorrelation function has a significant negative tail of a characteristic decay time equal to the mean time between collisions see the theoretical developments in Chapter 5 for details. 

Theory suggests that dips like the ones seen in Figs. 3.5, 3.31, and 3.34, may be modeled by an inverted Lorentzian profile, 

The weak dips near the rotovibrational H2 Si(l) line have been studied in some detail [316] and show the same dependence on density variation as those of the Q branch. For these, the parameter y of Eq. 3.14 is of course much smaller than unity. 

---